[js] annotation/metadatas parsing : improve, using masks ?

[GuillaumeDua]

TODO: refine such proposal

Current syntax, behavior

As is, the parser produce for the following input the code after :

• input

          THIS IS REMOVED // @awesome-doc-code-sections::skip::line
  auto main() -> int {    
                          // @awesome-doc-code-sections::show::block::begin
  auto i = 42; // test
  return i;
                          // @awesome-doc-code-sections::show::block::end
  }                       
          THIS IS REMOVED TOO // @awesome-doc-code-sections::skip::line
  

• output

  auto i = 42; // test
  return i;

Proposal

The syntax above is verbose and heavy, on purpose. However, this can be improve in multiple ways :

• use some filter logic

  • D aliases delete
  • S aliases show (kept)
  • H aliases hidden (but kept)

• use scopes annotation

• use default filter

  • // adcs:filter:default=D : everything is deleted by default
  • default's default is 'show'

• input

  // adcs:filter:default=H
  THIS IS REMOVED // @adcs:filter=D
  constexpr auto value = 42; // @adcs:filter=S
  auto main() -> int {    
  // @adcs:filter=S{
     auto i = value;
     return i;
  // @adcs:filter=}S
  }
  

• Output (show)

  constexpr auto value = 42;
     auto i = value;
     return i;

• output (execute)

  constexpr auto value = 42;
  auto main() -> int { 
     auto i = value;
     return i;
  }

Misc

• Decouples includes_transformations ?
• (optionaly) decouples metadata from the code ? (index+offset based descriptor)